[{"data":1,"prerenderedAt":1605},["ShallowReactive",2],{"site-nav-content":3,"blog:/blog/what-is-an-agentic-workflow":163,"blog-index-copy":651,"blog:/blog/what-is-an-agentic-workflow:surround":672,"hiring-banner-content":1574,"site-cta-content":1586},{"header":4,"productNav":9,"nav":42,"footer":61,"askAI":116,"id":147,"title":148,"archived":149,"authors":150,"badge":150,"body":151,"date":150,"department":150,"description":155,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":150,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":159,"relatedHeading":150,"seo":160,"series":150,"sitemap":115,"status":150,"stem":161,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":162},{"productLabel":5,"loginLabel":6,"contactLabel":7,"contactSalesLabel":8},"Product","Log in","Contact","Get started for free",[10,14,18,22,26,30,34,38],{"label":11,"to":12,"description":13},"Overview","/overview","Seven layers. One closed loop.",{"label":15,"to":16,"description":17},"Conflux","/product/conflux","Where your team, workstreams, and agents meet.",{"label":19,"to":20,"description":21},"Agent Teams","/product/agent-teams","Specialist teams - governed from day one.",{"label":23,"to":24,"description":25},"Lifecycle Graph","/product/lifecycle-graph","Intelligence that compounds across every interaction.",{"label":27,"to":28,"description":29},"Company Wiki","/product/wiki","Playbooks and policies where expertise stays.",{"label":31,"to":32,"description":33},"Workstreams","/product/workstreams","From brief to signed-off deliverable on one canvas.",{"label":35,"to":36,"description":37},"Perception Console","/product/perception","Ask your whole business in plain English.",{"label":39,"to":40,"description":41},"Governance","/product/governance","Frontier AI you can actually sign off on.",[43,46,49,52,55,58],{"label":44,"to":45},"Models","/models",{"label":47,"to":48},"Pricing","/pricing",{"label":50,"to":51},"Integrations","/integrations",{"label":53,"to":54},"Security","/security",{"label":56,"to":57},"Partners","/partners",{"label":59,"to":60},"Insights","/blog",{"productHeading":5,"companyHeading":62,"legalHeading":63,"docsLabel":64,"docsUrl":65,"statementLines":66,"copyright":69,"companyLinks":70,"legalLinks":85,"socialLinks":95,"bottomLinks":105},"Company","Legal","Docs","https://docs.gonimbus.ai",[67,68],"Stop training someone else's model.","Control your AI.","© 2026 Nimbus Intelligence, Inc. All rights reserved.",[71,72,73,74,75,77,80,83],{"label":47,"to":48},{"label":50,"to":51},{"label":53,"to":54},{"label":59,"to":60},{"label":76,"to":57},"Partner Program",{"label":78,"to":79},"Careers","/careers",{"label":81,"to":82},"System status","/status",{"label":7,"to":84},"/contact",[86,89,92],{"label":87,"to":88},"Terms of Service","/terms",{"label":90,"to":91},"Privacy Policy","/privacy",{"label":93,"to":94},"Compliance","/compliance",[96,99,102],{"label":97,"href":98},"LinkedIn","https://www.linkedin.com/company/gonimbusai/",{"label":100,"href":101},"X","https://x.com/gonimbusai",{"label":103,"href":104},"Instagram","https://www.instagram.com/gonimbus_ai/",[106,108,110,111,112],{"label":107,"to":88},"Terms",{"label":109,"to":91},"Privacy",{"label":93,"to":94},{"label":81,"to":82},{"label":113,"to":114,"external":115},"LLMs.txt","/llms.txt",true,{"text":117,"prompt":118},"Ask AI about Nimbus",{"I'm researching enterprise intelligence platforms and want to know how Nimbus combines perception, collaboration, and autonomous agents to drive strategic decision-making":119,"platforms":121},{" Summarize the highlights from Nimbus's website":120},"https://gonimbus.ai",[122,127,132,137,142],{"name":123,"label":124,"icon":125,"hrefPrefix":126},"chatgpt","ChatGPT","simple-icons:openai","https://chatgpt.com/?prompt=",{"name":128,"label":129,"icon":130,"hrefPrefix":131},"perplexity","Perplexity","mdi:magnify","https://www.perplexity.ai/search/new?q=",{"name":133,"label":134,"icon":135,"hrefPrefix":136},"grok","Grok","simple-icons:x","https://x.com/i/grok?text=",{"name":138,"label":139,"icon":140,"hrefPrefix":141},"claude","Claude","simple-icons:anthropic","https://claude.ai/new?q=",{"name":143,"label":144,"icon":145,"hrefPrefix":146},"google-ai","Google AI","simple-icons:google","https://www.google.com/search?udm=50&aep=11&q=","content/shared/nav.md","Site navigation",false,null,{"type":152,"value":153,"toc":154},"minimark",[],{"title":155,"searchDepth":156,"depth":156,"links":157},"",2,[],"md","/shared/nav",{"title":148,"description":155},"shared/nav","p6IsjEfjcrwkjwvGsPcuEkpoJSeUn2vJXPTxLWQfw9M",{"id":164,"title":165,"archived":149,"authors":166,"badge":169,"body":171,"date":640,"department":150,"description":641,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":150,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":642,"relatedHeading":150,"seo":643,"series":644,"sitemap":115,"status":150,"stem":645,"subhead":150,"tags":646,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":650},"content/blog/what-is-an-agentic-workflow.md","What is an Agentic Workflow",[167],{"name":168,"to":120},"Nimbus Research",{"label":170},"Explainer",{"type":152,"value":172,"toc":615},[173,177,189,199,202,207,210,237,240,290,302,306,309,316,319,355,358,363,369,381,387,393,409,413,419,425,431,437,447,450,454,461,468,480,484,488,491,495,498,502,505,509,512,516,521,525,528,532,535,539,547,551,554,558,564,568,573,577,580,584,592,596],[174,175,176],"p",{},"“We have an agent” often means a chat that never knows when to stop. Someone types a goal. The model keeps calling tools until the budget dies, or until a human closes the tab. There is no finish line. There is a conversation that looked busy.",[174,178,179,180,184,185,188],{},"A ",[181,182,183],"strong",{},"workflow"," has steps and a stop. An ",[181,186,187],{},"agentic workflow"," is a sequence of steps an AI can run toward a goal, with rules for when to stop — including a person who must approve before a live system changes.",[174,190,191,198],{},[192,193,197],"a",{"href":194,"rel":195},"https://www.anthropic.com/engineering/building-effective-agents",[196],"nofollow","Anthropic’s note on building effective agents"," makes the same cut: workflows with tools and stop conditions, not endless chat. The note is worth reading because it is honest about the boring parts — encoding the job, bounding the tools, and deciding what “done” means — rather than treating fluency as a process.",[174,200,201],{},"Older automation without models is brittle but auditable. Models without a workflow are flexible but unaccountable. An agentic workflow is the attempt to get both: language where the input is messy, and a finish line where the company needs one.",[203,204,206],"h2",{"id":205},"words-youll-hear","Words you’ll hear",[174,208,209],{},"Vendors collapse three different layers into the word “agentic”:",[211,212,213,220,226],"ul",{},[214,215,216,219],"li",{},[181,217,218],{},"Agentic capability."," The model can use tools, plan, and reflect. At work, this is “it can search Drive and draft a note.” It is not yet a job.",[214,221,222,225],{},[181,223,224],{},"Agentic workflow."," A designed sequence of those capabilities, with business stop conditions. This article is about this layer. At work, this is “extract, compare to the playbook, quote the CRM fields, wait for the named signer, write or refuse.”",[214,227,228,231,232,236],{},[181,229,230],{},"Agent platform."," Identity, connectors, tests, and governance around many workflows. At work, this is closer to an ",[192,233,235],{"href":234},"what-is-an-enterprise-ai-operating-system","enterprise AI operating system",".",[174,238,239],{},"Other terms:",[211,241,242,248,254,264,274,280],{},[214,243,244,247],{},[181,245,246],{},"Tool."," An action the AI can take: search files, query CRM, post a message. At work, a tool is a hand. Hands are not roles, and they are not stop conditions.",[214,249,250,253],{},[181,251,252],{},"Stop condition."," Budget hit, waiting on approval, error, empty result, human cancel. “The model says it is done” is a weak stop by itself.",[214,255,256,259,260,236],{},[181,257,258],{},"Write-back."," The AI is allowed to change a live system, not just draft. See ",[192,261,263],{"href":262},"what-is-write-back-governance","What is write-back governance",[214,265,266,269,270,236],{},[181,267,268],{},"Human wait."," A step in the sequence, not an interruption. See ",[192,271,273],{"href":272},"what-is-human-in-the-loop-ai","What is human-in-the-loop AI",[214,275,276,279],{},[181,277,278],{},"Version."," Which workflow definition ran. When policy changes, retrieval changes. Operators need to know which version ran last Tuesday.",[214,281,282,285,286,236],{},[181,283,284],{},"MCP."," A common plug so AI apps can use the same tools. Plumbing. It does not define your stops. See ",[192,287,289],{"href":288},"what-is-model-context-protocol","What is Model Context Protocol",[174,291,179,292,296,297,301],{},[192,293,295],{"href":294},"what-is-an-ai-workstream","workstream"," is the company object that ",[298,299,300],"em",{},"hosts"," the workflow: brief, connectors, people, budget, finish line. The workflow is the sequence. The workstream is the job folder. Mixing those two words is how demos skip isolation.",[203,303,305],{"id":304},"why-you-should-care","Why you should care",[174,307,308],{},"Capability demos look like workflows. They are not. A fluent plan is not a paused run waiting on approval, a failed run that did not retry a write, or a replay of which step ran.",[174,310,311,312,315],{},"It affects you if the job is ",[181,313,314],{},"multi-step, tool-using, and repeated"," — the opposite of one-off chat. Close checklists, renewal playbooks, and incident runbooks already have steps. Encode those. If the job is not written down, you will encode folklore and then fight the folklore.",[174,317,318],{},"Practical rules:",[211,320,321,327,333,339,345],{},[214,322,323,326],{},[181,324,325],{},"Read-heavy workflows"," can be long. They should still finish in an artefact with sources.",[214,328,329,332],{},[181,330,331],{},"Write-heavy workflows"," should be short after the quote: one payload, one gate, one execution, one record. Do not hide ten writes in a “cleanup agent.”",[214,334,335,338],{},[181,336,337],{},"Human wait is a step",", not an interruption.",[214,340,341,344],{},[181,342,343],{},"Version the workflow."," Policy and retrieval drift. Last Tuesday’s run needs a definition you can still open.",[214,346,347,350,351,236],{},[181,348,349],{},"Budget is a stop."," See ",[192,352,354],{"href":353},"what-is-ai-token-economics","What is AI token economics",[174,356,357],{},"A mega-agent with “figure it out” as the spec is not a workflow. It is a hope.",[359,360,362],"h3",{"id":361},"what-changes-by-role","What changes by role",[174,364,365,368],{},[181,366,367],{},"Finance."," Close and forecast jobs already have checklists. An agentic workflow that posts a journal without a stop at the named signer is not “agentic.” It is unattended posting. Finance also needs spend stops so a retry loop cannot become the month’s inference bill.",[174,370,371,374,375,380],{},[181,372,373],{},"Legal."," Customer-facing steps and anything that asserts a term need a gate before send. Air Canada’s chatbot invented a bereavement fare and the company was held to it — ",[192,376,379],{"href":377,"rel":378},"https://www.cbc.ca/news/canada/british-columbia/air-canada-chatbot-lawsuit-1.7116416",[196],"CBC’s report"," is the cautionary case for “the workflow ended at the message.” Legal also cares that the workflow version is reconstructable.",[174,382,383,386],{},[181,384,385],{},"Operations."," This is the native language: runbooks, queues, retries, and “do not proceed.” Ops should refuse workflows that cannot pause cleanly, cannot show which step failed, and cannot distinguish “waiting on a person” from “waiting on a tool error.”",[174,388,389,392],{},[181,390,391],{},"Go-to-market."," Renewal and hygiene jobs are repeated and tool-using. GTM should demand a short write path after the quote, not a weekend “cleanup” that touches hundreds of records behind one click. Time-to-approved-write is the metric, not time-to-first-plan.",[174,394,395,398,399,402,403,408],{},[181,396,397],{},"Security."," Tool belts are attack surface. Prompt injection that tricks a model into ",[298,400,401],{},"requesting"," a write should still die at a fail-closed gate. Importing every MCP helper into one workflow is how a demo becomes one actor with every production login. The ",[192,404,407],{"href":405,"rel":406},"https://genai.owasp.org/llm-top-10/",[196],"OWASP Top 10 for LLM applications"," treats tool use as a security topic for this reason.",[359,410,412],{"id":411},"what-people-get-wrong","What people get wrong",[174,414,415,418],{},[181,416,417],{},"Chat as workflow."," A conversation that looks busy has no durable instance, no version, and no gate.",[174,420,421,424],{},[181,422,423],{},"A checklist in a prompt."," A start. Without tools, a durable job, and a stop, it is still a prompt.",[174,426,427,430],{},[181,428,429],{},"Replacing a stable bot."," If the job is a scheduled export, older automation is the right tool. Agentic workflows help on messy documents. They are not a prestige upgrade for a cron job.",[174,432,433,436],{},[181,434,435],{},"Fully autonomous production."," Only for actions you would already automate without a model, plus logging. If you would not let a scheduled job do it, do not let an agent do it unattended.",[174,438,439,442,443,236],{},[181,440,441],{},"Multi-agent as a requirement."," A single tool-using agent can execute a workflow. Multiple agents help when duties already split. See ",[192,444,446],{"href":445},"what-is-multi-agent-ai","What is multi-agent AI",[174,448,449],{},"Good looks like: named steps, bounded tools, explicit stops (including human wait and budget), versioned definitions, read-only by default, fail-closed writes. Failure looks like a flagship model with every connector and a spec that says “be helpful.”",[203,451,453],{"id":452},"how-this-shows-up-in-nimbus","How this shows up in Nimbus",[174,455,456,457,236],{},"Nimbus’s delivery unit for operators is the ",[181,458,459],{},[192,460,295],{"href":294},[174,462,463,464,467],{},"The mapping in everyday terms: the brief is the goal; ",[192,465,466],{"href":445},"agent teams"," run the steps; connectors are the tools (default read-only); wiki is the playbook the steps must respect; governance is the wait/write stop; the Lifecycle Graph is the executed run. Model routing chooses the brain per step; it does not choose the stop.",[174,469,470,471,473,474,477,478,236],{},"See ",[192,472,31],{"href":32},", ",[192,475,476],{"href":20},"Agent teams",", and ",[192,479,39],{"href":40},[203,481,483],{"id":482},"questions-people-actually-ask","Questions people actually ask",[359,485,487],{"id":486},"is-a-checklist-in-a-prompt-an-agentic-workflow","Is a checklist in a prompt an agentic workflow?",[174,489,490],{},"It is a start. If there are no tools, no durable instance, and no gate, it is a prompt.",[359,492,494],{"id":493},"how-is-this-different-from-older-robotic-automation","How is this different from older robotic automation?",[174,496,497],{},"Older automation executes deterministic steps. Agentic workflows add language and planning. That helps on messy documents. It also means you need tests and human gates. Do not replace a stable bot with an agent if the job is still a scheduled export.",[359,499,501],{"id":500},"do-agentic-workflows-require-multiple-agents","Do agentic workflows require multiple agents?",[174,503,504],{},"No. A single tool-using agent can execute a workflow. Multiple agents help when duties already split in the organisation.",[359,506,508],{"id":507},"can-a-workflow-be-fully-autonomous-in-production","Can a workflow be fully autonomous in production?",[174,510,511],{},"Only for actions you would already automate without a model, plus logging.",[359,513,515],{"id":514},"where-do-tool-connection-standards-fit","Where do tool-connection standards fit?",[174,517,518,519,236],{},"A common plug so AI apps can use the same tools is plumbing. It does not define your stops or approvals. See ",[192,520,289],{"href":288},[359,522,524],{"id":523},"what-is-a-good-stop-condition-besides-the-model-is-done","What is a good stop condition besides “the model is done”?",[174,526,527],{},"Budget ceiling, empty retrieval, tool error, human cancel, and wait-for-named-signer. “Done” from the model is a suggestion. Encode the others.",[359,529,531],{"id":530},"how-long-should-a-write-heavy-workflow-be","How long should a write-heavy workflow be?",[174,533,534],{},"Short after the quote. One payload, one gate, one execution, one record. Length belongs in the read and compare steps, not in a bundle of hidden mutations.",[359,536,538],{"id":537},"how-do-we-version-a-workflow-when-the-wiki-changes","How do we version a workflow when the wiki changes?",[174,540,541,542,546],{},"Treat the playbook version as an input to the run. The ",[192,543,545],{"href":544},"what-is-a-lifecycle-graph","lifecycle graph"," should cite which wiki version the steps respected. Changing policy without recording which definition ran is how Tuesday becomes unexplained.",[359,548,550],{"id":549},"is-agentic-the-same-as-autonomous","Is “agentic” the same as “autonomous”?",[174,552,553],{},"No. Agentic means the model can plan and use tools. Autonomy is a policy about whether a person must still sign. Most production writes should not be autonomous.",[359,555,557],{"id":556},"can-we-import-every-available-tool-and-let-the-model-choose","Can we import every available tool and let the model choose?",[174,559,560,561,236],{},"That is a confused workflow. Least privilege applies to tools as much as to data. See ",[192,562,563],{"href":294},"What is an AI workstream",[359,565,567],{"id":566},"how-does-this-relate-to-human-in-the-loop","How does this relate to human-in-the-loop?",[174,569,570,571,236],{},"Human wait is a first-class step. If the person is only “on the loop” with a kill switch, you have a different design. See ",[192,572,273],{"href":272},[359,574,576],{"id":575},"will-a-better-model-remove-the-need-for-a-workflow","Will a better model remove the need for a workflow?",[174,578,579],{},"Stronger models plan more fluently. They still do not know your finish line, your signer, or your budget. Fluency without stops is a more expensive loop.",[203,581,583],{"id":582},"related-reading","Related reading",[174,585,586,589,590,236],{},[192,587,588],{"href":234},"What is an enterprise AI operating system"," and ",[192,591,446],{"href":445},[203,593,595],{"id":594},"sources","Sources",[211,597,598,604,610],{},[214,599,600],{},[192,601,603],{"href":194,"rel":602},[196],"Anthropic, Building effective agents",[214,605,606],{},[192,607,609],{"href":377,"rel":608},[196],"CBC, Air Canada chatbot lawsuit",[214,611,612],{},[192,613,407],{"href":405,"rel":614},[196],{"title":155,"searchDepth":156,"depth":156,"links":616},[617,618,623,624,638,639],{"id":205,"depth":156,"text":206},{"id":304,"depth":156,"text":305,"children":619},[620,622],{"id":361,"depth":621,"text":362},3,{"id":411,"depth":621,"text":412},{"id":452,"depth":156,"text":453},{"id":482,"depth":156,"text":483,"children":625},[626,627,628,629,630,631,632,633,634,635,636,637],{"id":486,"depth":621,"text":487},{"id":493,"depth":621,"text":494},{"id":500,"depth":621,"text":501},{"id":507,"depth":621,"text":508},{"id":514,"depth":621,"text":515},{"id":523,"depth":621,"text":524},{"id":530,"depth":621,"text":531},{"id":537,"depth":621,"text":538},{"id":549,"depth":621,"text":550},{"id":556,"depth":621,"text":557},{"id":566,"depth":621,"text":567},{"id":575,"depth":621,"text":576},{"id":582,"depth":156,"text":583},{"id":594,"depth":156,"text":595},"2026-08-17","An agentic workflow is a sequence of steps an AI can run toward a goal, with rules for when to stop — including a person who must approve before a live system changes.","/blog/what-is-an-agentic-workflow",{"title":165,"description":641},"explainer","blog/what-is-an-agentic-workflow",[644,647,648,649],"agentic-workflow","agents","workstreams","OD070bDpNh56Kq9pajGR0tbzXSES_KnS9029g0lELm8",{"hero":652,"id":654,"title":655,"archived":149,"authors":150,"badge":150,"body":656,"date":150,"department":150,"description":660,"extension":158,"eyebrow":661,"faqHeader":150,"faqs":150,"footerBand":662,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":60,"relatedHeading":668,"seo":669,"series":150,"sitemap":115,"status":150,"stem":670,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":671},{"filename":653},"u2221455217_Flat_design_of_a_futuristic_minimalist_landscape__5d589295-cdea-4ea9-a262-be766881accf_1.png","content/blog/index.md","Exploring the future of intelligence.",{"type":152,"value":657,"toc":658},[],{"title":155,"searchDepth":156,"depth":156,"links":659},[],"Deep dives into pre-cognitive intelligence, sentient enterprises, and the evolving landscape of AI-driven business transformation.","Latest Research",{"headline":663,"description":664,"primaryLabel":665,"primaryTo":666,"secondaryLabel":667,"secondaryTo":12},"Stay at the frontier.","Subscribe for product updates and new insights.","Subscribe","/newsletter","Explore the platform","More research",{"title":655,"description":660},"blog/index","eK1RCXdDW8nfLSyKRXGB1mJm9FAmhAO6GWXwNKOMNVE",[673,1121],{"id":674,"title":675,"archived":149,"authors":676,"badge":678,"body":679,"date":640,"department":150,"description":1113,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":150,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":1114,"relatedHeading":150,"seo":1115,"series":644,"sitemap":115,"status":150,"stem":1116,"subhead":150,"tags":1117,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1120},"content/blog/what-is-an-ai-workstream.md","What is an AI Workstream",[677],{"name":168,"to":120},{"label":170},{"type":152,"value":680,"toc":1089},[681,688,691,694,703,706,708,785,792,794,800,803,822,825,850,853,855,860,868,873,878,886,888,894,899,905,911,917,923,926,933,935,942,945,951,963,965,969,972,976,979,983,986,990,996,1000,1005,1009,1012,1016,1023,1027,1030,1034,1041,1045,1048,1052,1055,1059,1064,1066,1073,1075],[174,682,683,684,687],{},"An AI workstream is a ",[181,685,686],{},"shared workspace for one job",": a brief, the tools allowed, the people and AI on it, a budget, and a finish line.",[174,689,690],{},"A Slack channel with a bot is not a job. It is a room. Anyone can paste anything. The bot never knows when the work is done. Next quarter, nobody can say which systems were in play or who was allowed to change them.",[174,692,693],{},"If you cannot name the systems in scope and the approval policy on writes, you do not have a workstream. You have a conversation.",[174,695,696,697,702],{},"Software teams already learned this. Work lives in issues and tickets, not in unbounded chat. ",[192,698,701],{"href":699,"rel":700},"https://www.atlassian.com/agile/project-management/epics-stories-themes",[196],"Atlassian’s epics and stories"," are named packages with a boundary. AI operations are still catching up. The missing object is often the work package: a place where the job actually lives.",[174,704,705],{},"The analogy is not decoration. Tickets have a requester, a scope, an owner, and a closed state. Copilots have a thread. Threads do not archive cleanly, do not attach least-privilege connectors, and do not carry a named signer. When AI started touching live systems, the thread stopped being a sufficient container.",[203,707,206],{"id":205},[211,709,710,716,722,738,744,752,758,766,776],{},[214,711,712,715],{},[181,713,714],{},"Brief."," What this job is for, and what “done” means. At work, “Q3 regional discount hygiene” is a brief. “My stuff” is not.",[214,717,718,721],{},[181,719,720],{},"Connector."," A secure link to a live system (CRM, ERP, Drive). Attach what this job needs — not every system “just in case.” Default is read-only.",[214,723,724,727,728,731,732,737],{},[181,725,726],{},"Scope / least privilege."," Only the data and tools required for ",[298,729,730],{},"this"," job. ",[192,733,736],{"href":734,"rel":735},"https://www.hhs.gov/hipaa/for-professionals/privacy/guidance/minimum-necessary-requirement/index.html",[196],"HIPAA’s minimum necessary"," is the same instinct: do not attach every system to every task. GDPR purpose limitation is the privacy-law cousin.",[214,739,740,743],{},[181,741,742],{},"God workspace."," One org-wide chat that can see every folder and every CRM object because setup was easier. At work, this is how recruiting sees finance forecasts.",[214,745,746,749,750,236],{},[181,747,748],{},"Agent team."," The AI specialists assigned to the job. The workstream is the stage; the team is the cast. See ",[192,751,446],{"href":445},[214,753,754,757],{},[181,755,756],{},"Named signer."," Who must approve a write. At work, this is a role that already owns that class of change.",[214,759,760,763,764,236],{},[181,761,762],{},"NTU / budget."," The spend ceiling for the job. See ",[192,765,354],{"href":353},[214,767,768,771,772,236],{},[181,769,770],{},"Wiki section."," The asserted playbooks this job may load. See ",[192,773,775],{"href":774},"what-is-a-company-wiki-for-ai-agents","What is a company wiki for AI agents",[214,777,778,781,782,236],{},[181,779,780],{},"Lifecycle Graph."," The chain this job emits as it runs. See ",[192,783,784],{"href":544},"What is a lifecycle graph",[174,786,787,788,791],{},"An ",[192,789,187],{"href":790},"what-is-an-agentic-workflow"," is the sequence of steps. The workstream is the durable instance those steps run inside. A workflow definition without a workstream is a script on someone’s laptop. A workstream without a workflow is a folder with no process.",[203,793,305],{"id":304},[174,795,796,797,799],{},"Without a boundary, two departments sharing an AI tool will either over-share (the recruiting job can see finance forecasts) or under-share (people export spreadsheets to personal ChatGPT). The workstream is the compromise: enough context to do ",[298,798,730],{}," job, not the whole company.",[174,801,802],{},"It affects you if work:",[211,804,805,808,811,814,817],{},[214,806,807],{},"touches more than one system",[214,809,810],{},"involves more than one role",[214,812,813],{},"can change a live record",[214,815,816],{},"needs a budget you can attribute",[214,818,819,820],{},"must still be explainable after people leave — see ",[192,821,784],{"href":544},[174,823,824],{},"Open workstreams the way you would open a ticket:",[211,826,827,834,841,844,847],{},[214,828,829,830,833],{},"One workstream per ",[181,831,832],{},"outcome",", not per person. “Q3 regional discount hygiene” can have several humans. “My stuff” cannot be governed or archived.",[214,835,836,837,840],{},"Attach the ",[181,838,839],{},"minimum"," connectors.",[214,842,843],{},"Set the write policy on day one, even if you start read-only.",[214,845,846],{},"Reuse templates, not last month’s chat thread.",[214,848,849],{},"Close or archive when the job ends. A sprint that never ends is not a sprint.",[174,851,852],{},"A standing “Ask AI” workstream with org-wide connectors recreates the copilot, including the blast radius.",[359,854,362],{"id":361},[174,856,857,859],{},[181,858,367],{}," Chargeback becomes possible because the job is named. Close workstreams can attach ERP read-only, load the close checklist from the wiki, and keep GTM out of the ledger. A company-wide AI pool with no workstream attribution is a shared kitchen.",[174,861,862,864,865,867],{},[181,863,373],{}," Scope is a processing purpose. A workstream for a renewal can include legal and go-to-market on ",[298,866,730],{}," goal without merging their entire universes. Legal also gets a closed state: when the job ends, retention follows the type of record instead of an immortal channel.",[174,869,870,872],{},[181,871,385],{}," This is the ticket analogue they already wanted. Ops should refuse god workspaces, insist on a finish line, and treat human wait as a status, not a side conversation in Slack.",[174,874,875,877],{},[181,876,391],{}," Cross-functional launches finally have a place that is not a merged Slack. GTM still should not get finance’s ERP “for context.” Templates beat copying last quarter’s thread, which silently copies last quarter’s over-attached connectors.",[174,879,880,882,883,885],{},[181,881,397],{}," Least privilege is now a product object, not a memo. Connectors default to read-only. Adding a write path is a deliberate change to ",[298,884,730],{}," job, not a tenant-wide toggle. A workstream that never closes is a standing access grant.",[359,887,412],{"id":411},[174,889,890,893],{},[181,891,892],{},"One workstream per person."," You cannot archive “my stuff.” You cannot attribute it. You cannot apply least privilege.",[174,895,896,898],{},[181,897,742],{}," Setup is easier. Blast radius is the company.",[174,900,901,904],{},[181,902,903],{},"ChatGPT Project as the unit."," Some files, some instructions. Typically no connector-level least privilege, quoted writes, spend caps, or lasting record. Fine for personal research. Not an operations unit.",[174,906,907,910],{},[181,908,909],{},"Too small."," If setup exceeds the job, use a lighter sanctioned copilot path. Do not open a workstream to rewrite one sentence.",[174,912,913,916],{},[181,914,915],{},"Too large."," If you cannot explain the purpose in one sentence, or you keep attaching “one more connector,” split.",[174,918,919,922],{},[181,920,921],{},"Never closing."," Standing rooms recreate Slack, including the archaeology problem.",[174,924,925],{},"Good looks like: one outcome, minimum connectors, write policy on day one, wiki sections subscribed, budget capped, named signer, archive when done. Failure looks like an org-wide copilot with every OAuth grant and a channel that outlives the campaign.",[174,927,928,929,932],{},"The ",[192,930,931],{"href":234},"enterprise AI OS"," metaphor is isolation plus I/O plus state. The workstream is the isolation unit. Without it, connectors, wiki, and agent teams have nowhere to attach that an auditor could name.",[203,934,453],{"id":452},[174,936,937,938,941],{},"In Nimbus, workstreams are how ",[192,939,940],{"href":790},"agentic workflows"," become company objects rather than a file only one engineer can run.",[174,943,944],{},"Each workstream carries a brief, wiki sections (approved playbooks), connector attachments (read-only by default), agent team assignment, spend budget, release policy on writes, and nodes on the Lifecycle Graph.",[174,946,947,948,950],{},"Cross-department work is multiple teams on one workstream, not a merged Slack. Operators open this themselves; the point of an ",[192,949,235],{"href":234}," is that the job folder is a product, not a forward-deployed spreadsheet.",[174,952,470,953,955,956,473,958,473,961,236],{},[192,954,31],{"href":32},". Related product: ",[192,957,476],{"href":20},[192,959,960],{"href":28},"Wiki",[192,962,39],{"href":40},[203,964,483],{"id":482},[359,966,968],{"id":967},"is-a-chatgpt-project-a-workstream","Is a ChatGPT “Project” a workstream?",[174,970,971],{},"It is a weak analogue: some files, some custom instructions. It typically lacks connector-level least privilege, quoted writes, spend caps, and a lasting record. Useful for personal research. Not an operations unit.",[359,973,975],{"id":974},"how-small-is-too-small","How small is too small?",[174,977,978],{},"If the setup cost exceeds the job, use a lighter sanctioned copilot path. Do not create a workstream to rewrite one sentence.",[359,980,982],{"id":981},"how-large-is-too-large","How large is too large?",[174,984,985],{},"If you cannot explain the purpose in one sentence, or you keep attaching “one more connector,” split.",[359,987,989],{"id":988},"can-one-workstream-serve-multiple-departments","Can one workstream serve multiple departments?",[174,991,992,993,995],{},"Yes — go-to-market and legal on a renewal, for example. They share ",[298,994,730],{}," goal’s scope, not each other’s entire universe.",[359,997,999],{"id":998},"how-do-we-budget-them","How do we budget them?",[174,1001,1002,1003,236],{},"Caps per workstream, plus an organisation pool. Chargeback by workstream beats “the AI bill.” See ",[192,1004,354],{"href":353},[359,1006,1008],{"id":1007},"is-a-slack-channel-with-a-bot-enough-if-we-add-a-approve-command","Is a Slack channel with a bot enough if we add a /approve command?",[174,1010,1011],{},"No. A command is not connector least privilege, a quoted payload, a durable chain, or an archive policy. It is still a room.",[359,1013,1015],{"id":1014},"who-is-allowed-to-open-a-workstream","Who is allowed to open a workstream?",[174,1017,1018,1019,236],{},"Whoever is allowed to open that class of job in analogue life — with the same instinct as who may open a ticket or a change request. An “AI team” bottleneck recreates the waitlist that causes ",[192,1020,1022],{"href":1021},"what-is-shadow-ai","shadow AI",[359,1024,1026],{"id":1025},"what-happens-when-the-job-ends","What happens when the job ends?",[174,1028,1029],{},"Close or archive. Revoke standing connector usefulness. Keep the reconstructable chain according to retention, not the entire chat.",[359,1031,1033],{"id":1032},"do-we-need-a-workstream-for-read-only-analysis","Do we need a workstream for read-only analysis?",[174,1035,1036,1037,236],{},"When the analysis crosses systems, roles, or must be replayed later, yes. When it is personal drafting with no live-system scope, a sanctioned copilot may be enough. See ",[192,1038,1040],{"href":1039},"how-to-choose-between-a-copilot-and-a-work-os","How to choose between a copilot and a work OS",[359,1042,1044],{"id":1043},"how-do-wiki-and-connectors-differ-inside-a-workstream","How do wiki and connectors differ inside a workstream?",[174,1046,1047],{},"Wiki is asserted policy the job must follow. Connectors are live systems the job may read (and, if enabled, write). Mixing them into one “knowledge” pile is how Drive folklore overwrites the playbook.",[359,1049,1051],{"id":1050},"can-we-keep-one-standing-workstream-for-ask-anything","Can we keep one standing workstream for “ask anything”?",[174,1053,1054],{},"You can. You will recreate the copilot, including over-share. Standing Q&A belongs on a tightly scoped, read-only path if it exists at all.",[359,1056,1058],{"id":1057},"how-does-this-relate-to-agent-teams","How does this relate to agent teams?",[174,1060,1061,1062,236],{},"The workstream is the job. The agent team is the cast assigned to it. Changing the cast does not change the brief, the connectors, or the signer. See ",[192,1063,446],{"href":445},[203,1065,583],{"id":582},[174,1067,1068,589,1070,236],{},[192,1069,588],{"href":234},[192,1071,1072],{"href":790},"What is an agentic workflow",[203,1074,595],{"id":594},[211,1076,1077,1083],{},[214,1078,1079],{},[192,1080,1082],{"href":699,"rel":1081},[196],"Atlassian, epics, stories, and themes",[214,1084,1085],{},[192,1086,1088],{"href":734,"rel":1087},[196],"HHS, HIPAA minimum necessary requirement",{"title":155,"searchDepth":156,"depth":156,"links":1090},[1091,1092,1096,1097,1111,1112],{"id":205,"depth":156,"text":206},{"id":304,"depth":156,"text":305,"children":1093},[1094,1095],{"id":361,"depth":621,"text":362},{"id":411,"depth":621,"text":412},{"id":452,"depth":156,"text":453},{"id":482,"depth":156,"text":483,"children":1098},[1099,1100,1101,1102,1103,1104,1105,1106,1107,1108,1109,1110],{"id":967,"depth":621,"text":968},{"id":974,"depth":621,"text":975},{"id":981,"depth":621,"text":982},{"id":988,"depth":621,"text":989},{"id":998,"depth":621,"text":999},{"id":1007,"depth":621,"text":1008},{"id":1014,"depth":621,"text":1015},{"id":1025,"depth":621,"text":1026},{"id":1032,"depth":621,"text":1033},{"id":1043,"depth":621,"text":1044},{"id":1050,"depth":621,"text":1051},{"id":1057,"depth":621,"text":1058},{"id":582,"depth":156,"text":583},{"id":594,"depth":156,"text":595},"An AI workstream is a shared workspace for one job: a brief, the tools allowed, the people and AI on it, a budget, and a finish line — not a Slack channel with a bot.","/blog/what-is-an-ai-workstream",{"title":675,"description":1113},"blog/what-is-an-ai-workstream",[644,649,1118,1119],"enterprise-ai","governance","Auw7pxXLBHz5skUcPAsGCeasYr2BpxrUjMrBsoRzRYo",{"id":1122,"title":1123,"archived":149,"authors":1124,"badge":1126,"body":1127,"date":640,"department":150,"description":1565,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":150,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":1566,"relatedHeading":150,"seo":1567,"series":644,"sitemap":115,"status":150,"stem":1568,"subhead":150,"tags":1569,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1573},"content/blog/what-is-ai-token-economics.md","What is AI Token Economics",[1125],{"name":168,"to":120},{"label":170},{"type":152,"value":1128,"toc":1541},[1129,1146,1152,1155,1162,1164,1248,1259,1261,1264,1284,1287,1310,1313,1315,1324,1329,1334,1339,1347,1349,1355,1365,1371,1379,1389,1398,1400,1403,1409,1418,1420,1424,1427,1431,1434,1438,1441,1445,1450,1454,1457,1461,1464,1468,1473,1477,1480,1484,1487,1491,1496,1500,1505,1509,1512,1514,1520,1522],[174,1130,179,1131,1134,1135,589,1140,1145],{},[181,1132,1133],{},"token"," is a chunk of text the model reads or writes. You pay per chunk. Different models cost different amounts. Input, output, and sometimes tools all meter differently. ",[192,1136,1139],{"href":1137,"rel":1138},"https://openai.com/api/pricing/",[196],"OpenAI",[192,1141,1144],{"href":1142,"rel":1143},"https://www.anthropic.com/pricing",[196],"Anthropic"," publish those ladders. Finance still cannot run the business on “12 million tokens of vendor A’s flagship.”",[174,1147,1148,1151],{},[181,1149,1150],{},"AI token economics"," is treating that usage like a real budget: measuring, allocating, controlling, and attributing spend so operators can quote before a run, cap during it, and attribute after it — instead of a slide that says “unlimited AI.”",[174,1153,1154],{},"Without it, organisations either freeze (no production AI) or send every small task to the most expensive model until the bill becomes a board slide.",[174,1156,1157,1158,1161],{},"The unit problem is the same one cloud had in its first decade: a metered resource sold with a headcount story. Seat licences predict people. Inference predicts work. When those two are collapsed into “unlimited,” the next chunk ",[298,1159,1160],{},"feels"," free, so people pick the flagship every time. The ladder did not disappear. It hid.",[203,1163,206],{"id":205},[211,1165,1166,1172,1178,1187,1197,1203,1209,1218,1224,1235,1241],{},[214,1167,1168,1171],{},[181,1169,1170],{},"Token."," A piece of text the model processes. Not a business unit. At work, a long wiki dump and a short field extract are wildly different token counts for the same “question.”",[214,1173,1174,1177],{},[181,1175,1176],{},"Seat licence."," Predictable cost per person. Often marketed as “unlimited.” The underlying work is still metered.",[214,1179,1180,1183,1184,1186],{},[181,1181,1182],{},"Pass-through API bill."," Each team has keys. Simple. Invites key sprawl and ",[192,1185,1022],{"href":1021}," on personal keys. At work, the invoice lands in engineering while go-to-market did the looping.",[214,1188,1189,1192,1193,1196],{},[181,1190,1191],{},"Quote."," A number ",[298,1194,1195],{},"before"," they run. At work, this is what makes a brief a decision rather than a surprise.",[214,1198,1199,1202],{},[181,1200,1201],{},"Cap / ceiling."," A hard stop. The loop cannot spend past it. At work, weekend agent loops die here instead of in next month’s cloud bill.",[214,1204,1205,1208],{},[181,1206,1207],{},"Pool."," Organisation-level allowance. At work, one department should not be able to burn the company pool on a vanity run.",[214,1210,1211,1214,1215,1217],{},[181,1212,1213],{},"Attribution."," Chargeback by job, not “the AI bill.” At work, finance can ask which ",[192,1216,295],{"href":294}," consumed the units.",[214,1219,1220,1223],{},[181,1221,1222],{},"NTU (Nimbus Token Unit)."," Nimbus’s normalised work credit for completed AI activity — analysis, tools, runs, writes — sitting above raw provider tokens. Everyday questions can be included; heavier work consumes pool credits. Finance gets one tape measure across vendors and steps.",[214,1225,1226,1229,1230,1234],{},[181,1227,1228],{},"Model routing."," Cheaper model for simple steps, stronger only when needed. See ",[192,1231,1233],{"href":1232},"what-is-model-routing","What is model routing",". At work, classify-this-ticket should not pay flagship rates.",[214,1236,1237,1240],{},[181,1238,1239],{},"Context window."," How much text the model can see at once. Dumping the whole Drive into context is an economic choice, not a quality strategy.",[214,1242,1243,1245,1246,236],{},[181,1244,252],{}," Budget hit, empty result, human cancel. Agent loops can dominate the bill without improving the artefact. See ",[192,1247,1072],{"href":790},[174,1249,1250,1251,1254,1255,1258],{},"The point is ",[181,1252,1253],{},"value per unit",", not minimum units regardless of outcome. Caching, wiki citations, and memory should make the ",[298,1256,1257],{},"same"," outcome cheaper over time. If unit cost of an approved update never falls, you are re-deriving folklore every run.",[203,1260,305],{"id":304},[174,1262,1263],{},"It affects you if you:",[211,1265,1266,1272,1278],{},[214,1267,1268,1271],{},[181,1269,1270],{},"Own the budget."," Surprise invoices arrive after agents looped all weekend.",[214,1273,1274,1277],{},[181,1275,1276],{},"Run the work."," You should see a number before you commit, not a lecture after.",[214,1279,1280,1283],{},[181,1281,1282],{},"Are tempted to shame people for using AI."," Shame drives personal keys. Cap the official path so it is safe to use.",[174,1285,1286],{},"Practical rhythm:",[211,1288,1289,1298,1304],{},[214,1290,1291,1294,1295,1297],{},[181,1292,1293],{},"Name the run."," Unnamed chats cannot be attributed. That is what a ",[192,1296,295],{"href":294}," is for.",[214,1299,1300,1303],{},[181,1301,1302],{},"Separate exploration from production."," Sandboxes can have tighter caps and cheaper default routes.",[214,1305,1306,1309],{},[181,1307,1308],{},"Review unit cost of outcomes"," — approved updates per unit — not tokens in the abstract.",[174,1311,1312],{},"Anti-pattern: a single corporate API key in a wiki, no per-job cap, monthly surprise. That is an unmetered utility.",[359,1314,362],{"id":361},[174,1316,1317,1319,1320,1323],{},[181,1318,367],{}," You need a quote, a ceiling, and a chargeback dimension that matches how the business already thinks — by job, department, or cost centre — not by vendor token type. Multi-vendor ladders are incomparable until you normalise. NTU is that normalisation in Nimbus. Finance should also see ",[298,1321,1322],{},"stops",": a cap that fired is a successful control, not a failed project.",[174,1325,1326,1328],{},[181,1327,373],{}," Spend logs are not only money. They are a map of which data classes went to which provider. Uncapped personal keys are a processing-agreement gap. Legal will also ask whether you can stop a run, not only whether you can pay for it.",[174,1330,1331,1333],{},[181,1332,385],{}," Caps are operational stops, like a queue limit. Ops needs to know whether a paused run is waiting on a person or waiting on budget. Mixing those two in one “it failed” status is how you get the wrong pager.",[174,1335,1336,1338],{},[181,1337,391],{}," GTM feels the quality-versus-cost trade first. A compact model that extracts fields is usually enough. A flagship model that argues a clause may be worth it. Without routing and quotes, GTM either hoards “the best model” or gets blamed for the bill. Neither produces better pipeline hygiene.",[174,1340,1341,1343,1344,1346],{},[181,1342,397],{}," API keys are credentials. Personal keys in browser plugins are ",[192,1345,1022],{"href":1021},". A pooled official path with per-workstream ceilings reduces key sprawl. Spend spikes can also be an anomaly signal — a loop that never stops is sometimes a bug, sometimes a prompt-injection success.",[359,1348,412],{"id":411},[174,1350,1351,1354],{},[181,1352,1353],{},"“Unlimited” as a strategy."," Seats hide the ladder. They do not delete it. Heavy agentic work will still surface as a true-up, a throttle, or a degraded model.",[174,1356,1357,1360,1361,1364],{},[181,1358,1359],{},"Punishing usage."," Chargeback without a sanctioned path recreates personal keys. Celebrate lower units ",[298,1362,1363],{},"per artefact"," as playbooks and memory compound.",[174,1366,1367,1370],{},[181,1368,1369],{},"Tokens as the KPI."," Tokens measure consumption. Outcomes measure value. A cheap run that produces a rejected write is still waste. A dearer run that produces one approved journal may be fine.",[174,1372,1373,1376,1377,236],{},[181,1374,1375],{},"One model for everything."," That is a routing failure dressed as quality culture. See ",[192,1378,1233],{"href":1232},[174,1380,1381,1384,1385,1388],{},[181,1382,1383],{},"No stop on loops."," ",[192,1386,197],{"href":194,"rel":1387},[196]," treats workflows with stop conditions as the grown-up shape. Economics is one of those stops.",[174,1390,1391,1392,589,1395,1397],{},"Good looks like: named jobs, quotes before commit, hard ceilings, routing policy, attribution, and falling unit cost as the ",[192,1393,1394],{"href":774},"wiki",[192,1396,545],{"href":544}," reduce re-derivation. Failure looks like a shared key, a flagship default, and a board slide titled “AI spend.”",[203,1399,453],{"id":452},[174,1401,1402],{},"Workstreams show quotes and ceilings before runs. Orgs draw from a pooled NTU allowance. Routing is a policy, not a dropdown labelled “best.” Memory and wiki reduce re-derivation, which is how unit cost of an outcome should fall over time.",[174,1404,1405,1406,1408],{},"Everyday questions can sit inside the allowance; heavier analysis, tools, and writes consume pool credits. The ",[192,1407,23],{"href":544}," can record spend as part of the chain, so “the run stopped because the ceiling was hit” is a causal fact.",[174,1410,470,1411,589,1413,1415,1416,236],{},[192,1412,44],{"href":45},[192,1414,1233],{"href":1232},". Product context: ",[192,1417,31],{"href":32},[203,1419,483],{"id":482},[359,1421,1423],{"id":1422},"why-cant-we-just-pay-seats-and-call-it-unlimited","Why can’t we just pay seats and call it unlimited?",[174,1425,1426],{},"Seats predict headcount. Production AI spend is inference, tools, and writes. “Unlimited” hides the ladder; it does not delete it.",[359,1428,1430],{"id":1429},"what-should-finance-actually-see","What should finance actually see?",[174,1432,1433],{},"A quote before commit, a cap during the run, and attribution by job afterwards — in one unit they can compare across vendors and steps.",[359,1435,1437],{"id":1436},"wont-cheaper-models-get-worse-answers","Won’t cheaper models get worse answers?",[174,1439,1440],{},"For extract and classify, often no. For hard judgment, often yes. That is a routing policy, not a religion. Measure reject rates on the job, not vibes.",[359,1442,1444],{"id":1443},"do-we-punish-teams-for-using-ai","Do we punish teams for using AI?",[174,1446,1447,1448,1364],{},"No. Punishing usage revives shadow AI. Celebrate lower units ",[298,1449,1363],{},[359,1451,1453],{"id":1452},"what-is-an-ntu-in-plain-language","What is an NTU in plain language?",[174,1455,1456],{},"A normalised work credit above raw provider tokens, so a finance partner is not asked to compare “vendor A input tokens” with “vendor B output tokens” plus tool calls. In Nimbus, completed activity — analysis, tools, runs, writes — is what consumes the unit.",[359,1458,1460],{"id":1459},"should-every-chat-be-billed-to-a-cost-centre","Should every chat be billed to a cost centre?",[174,1462,1463],{},"Named production jobs, yes. Tiny sanctioned copilots for personal drafting can live on a lighter path. The failure is mixing them so neither can be capped.",[359,1465,1467],{"id":1466},"how-do-agent-loops-blow-the-budget","How do agent loops blow the budget?",[174,1469,1470,1471,236],{},"They call tools, re-read context, and retry without a finish line. Without a ceiling and a stop condition, “being thorough” is an unbounded loop. See ",[192,1472,1072],{"href":790},[359,1474,1476],{"id":1475},"is-caching-the-same-as-token-economics","Is caching the same as token economics?",[174,1478,1479],{},"Caching is a tactic. Economics is the management system: quote, cap, attribute, route. Caching without attribution still leaves you unable to explain the bill.",[359,1481,1483],{"id":1482},"do-we-need-a-data-warehouse-to-do-this","Do we need a data warehouse to do this?",[174,1485,1486],{},"You need events at run time. A warehouse can hold copies for reporting. It cannot quote a run that has not emitted a number yet.",[359,1488,1490],{"id":1489},"how-does-this-relate-to-write-back","How does this relate to write-back?",[174,1492,1493,1494,236],{},"Writes are usually a small number of tokens and a large operational risk. Do not use spend as a substitute for a named signer. Do use spend as a stop so a looping agent cannot keep proposing writes all weekend. See ",[192,1495,263],{"href":262},[359,1497,1499],{"id":1498},"can-we-lock-one-vendor-to-simplify-pricing","Can we lock one vendor to simplify pricing?",[174,1501,1502,1503,236],{},"You can. You will pay for it in price, outages, and lock-in. A normalised unit plus routing is how finance keeps a second tape measure. See ",[192,1504,1233],{"href":1232},[359,1506,1508],{"id":1507},"why-not-just-set-a-monthly-company-cap","Why not just set a monthly company cap?",[174,1510,1511],{},"A company cap without per-job attribution is a shared kitchen. The loudest workflow starves the others, and nobody can say which job did it.",[203,1513,583],{"id":582},[174,1515,1516,589,1518,236],{},[192,1517,1233],{"href":1232},[192,1519,563],{"href":294},[203,1521,595],{"id":594},[211,1523,1524,1530,1536],{},[214,1525,1526],{},[192,1527,1529],{"href":1137,"rel":1528},[196],"OpenAI API pricing",[214,1531,1532],{},[192,1533,1535],{"href":1142,"rel":1534},[196],"Anthropic pricing",[214,1537,1538],{},[192,1539,603],{"href":194,"rel":1540},[196],{"title":155,"searchDepth":156,"depth":156,"links":1542},[1543,1544,1548,1549,1563,1564],{"id":205,"depth":156,"text":206},{"id":304,"depth":156,"text":305,"children":1545},[1546,1547],{"id":361,"depth":621,"text":362},{"id":411,"depth":621,"text":412},{"id":452,"depth":156,"text":453},{"id":482,"depth":156,"text":483,"children":1550},[1551,1552,1553,1554,1555,1556,1557,1558,1559,1560,1561,1562],{"id":1422,"depth":621,"text":1423},{"id":1429,"depth":621,"text":1430},{"id":1436,"depth":621,"text":1437},{"id":1443,"depth":621,"text":1444},{"id":1452,"depth":621,"text":1453},{"id":1459,"depth":621,"text":1460},{"id":1466,"depth":621,"text":1467},{"id":1475,"depth":621,"text":1476},{"id":1482,"depth":621,"text":1483},{"id":1489,"depth":621,"text":1490},{"id":1498,"depth":621,"text":1499},{"id":1507,"depth":621,"text":1508},{"id":582,"depth":156,"text":583},{"id":594,"depth":156,"text":595},"AI token economics is treating AI usage like a real budget: you pay per chunk of text the model reads and writes, so finance can quote, cap, and attribute spend instead of hoping for “unlimited AI.”","/blog/what-is-ai-token-economics",{"title":1123,"description":1565},"blog/what-is-ai-token-economics",[644,1570,1571,1572],"token-economics","ntu","model-routing","kgDCkz6CVtDSpJc699QfUH1nMlvaVezkE2PrGzu2Puc",{"enabled":149,"message":1575,"linkLabel":78,"linkHref":79,"id":1576,"title":1577,"archived":149,"authors":150,"badge":150,"body":1578,"date":150,"department":150,"description":155,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":150,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":1582,"relatedHeading":150,"seo":1583,"series":150,"sitemap":115,"status":150,"stem":1584,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1585},"We're hiring! Join the team building the Sentient Enterprise.","content/shared/hiring.md","Hiring banner",{"type":152,"value":1579,"toc":1580},[],{"title":155,"searchDepth":156,"depth":156,"links":1581},[],"/shared/hiring",{"title":1577,"description":155},"shared/hiring","-6bioYD7lKYokGUVU3ff4hHTvB-sDyOMuCptKHnojfk",{"fold":1587,"id":1591,"title":1592,"archived":149,"authors":150,"badge":150,"body":1593,"date":150,"department":150,"description":155,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":1597,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":1601,"relatedHeading":150,"seo":1602,"series":150,"sitemap":115,"status":150,"stem":1603,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1604},{"headline":1588,"description":1589,"primaryLabel":8,"primaryTo":1590,"secondaryLabel":667,"secondaryTo":12},"Run frontier AI your business actually owns.","Governed agent swarms, 2,000+ integrations, and a knowledge graph that stays inside your walls. Free 7-day trial.","/checkout","content/shared/cta.md","Site CTAs",{"type":152,"value":1594,"toc":1595},[],{"title":155,"searchDepth":156,"depth":156,"links":1596},[],{"headline":1598,"description":1599,"primaryLabel":8,"primaryTo":1590,"secondaryLabel":1600,"secondaryTo":84},"See what governed AI looks like on your stack.","Connect your tools, run a workstream, and keep every decision on your ledger - free for 7 days.","Talk to our team","/shared/cta",{"title":1592,"description":155},"shared/cta","YHK6Fb8AvCPR1zZq7R_xiXUG0hwhP5UxHA8Ix52JQp4",1787194072928]